论文

RPT:基于语义引导掩码查询的雷达全景分割模型

Radar Panoptic Transformer: Semantic-Guided Mask Queries for Dynamic Scene Understanding

精选理由

车载雷达点云稀疏,动态目标常常只剩几个点,分割一直很难做。这篇论文的 RPT 用语义预测来初始化掩码查询,跳过了几何聚类,在 RadarScenes 拿到 71.6% mIoU,还能 31Hz 实时跑。做自动驾驶感知、尤其是低成本雷达方案的同学,值得仔细看。

论文提出 Radar Panoptic Transformer(RPT),一个端到端的点云网络,解决车载雷达点云稀疏、动态目标检测少导致的实例分割难题。模型通过分组向量注意力聚合位姿对齐的时序上下文,并根据距离、方位角和雷达散射截面(RCS)自适应学习每个点的邻域半径。其核心机制是用逐点语义预测在疑似前景位置初始化查询,再通过查询-点类别一致性引导掩码交叉注意力,直接生成实例掩码,无需几何聚类,并引入 RCS 一致性辅助损失对齐实例级统计。

正文

在基准成绩上,RPT 在 RadarScenes 上取得 71.6% mIoU 和 82.2% PQ 的最优表现,并以 31Hz 实现实时端到端推理,超过传感器帧率。在 230 米量程的 MAN TruckScenes 全向远距雷达场景中,RPT 也超过被评估的基线方法。

对自动驾驶研发者而言,该工作表明雷达可以独立承担全景分割任务,不再依赖摄像头或激光雷达点云,适合低速成本、恶劣天气下的感知方案。31Hz 的推理速度满足车载实时部署要求,远距离 230 米的表现对卡车和高速场景有直接参考价值。

原文 · arXiv cs.LG

Radar Panoptic Transformer: Semantic-Guided Mask Queries for Dynamic Scene Understanding

Automotive radar provides long-range sensing and direct Doppler velocity measurements, but its sparse and non-uniformly sampled point clouds pose significant challenges for panoptic segmentation. Dynamic objects often yield only a few detections amid a dominant static background, providing limiting the geometric cues for separating object instances. We propose \emph{Radar Panoptic Transformer (RPT)}, an end-to-end point-based network that incorporates radar-adaptive feature extraction with semantic-guided mask prediction. Its backbone aggregates pose-aligned temporal context through grouped vector attention and learns per-point neighborhood radii conditioned on range, azimuth, and radar cross section (RCS). Point-wise semantic predictions initialize queries at likely foreground detections and guide masked cross-attention through query-point class agreement, producing instance masks without geometric clustering. An auxiliary RCS-consistency loss matches instance-level statistics between predicted and annotated instances. RPT achieves state-of-the-art performance on RadarScenes with $71.6\%$ mIoU and $82.2\%$ PQ. We further evaluate RPT on MAN TruckScenes, where it outperforms the evaluated baselines in a long-range, full-surround radar setting with sensing ranges up to $230\,\mathrm{m}$. RPT achieves real-time end-to-end inference at $31$\,Hz on RadarScenes, exceeding the sensor frame rate.

产业坐标与信源核验

本条情报归属 AITOP「论文」核心追踪栏目。在跨区域产业观察坐标中,该动态直接关联「AI 开发者工具链、Coding Agent 与工程化实践」研究主线,反映出当前技术演进与商业落地的最新风向。涉及核心实体与技术议题包括:RadarScenes、TruckScenes、全景分割、自动驾驶,读者可通过下方标签进入对应专题页查阅历史脉络与关联对比。

信源核验说明:本文原始线索采编自「arXiv cs.LG」(@Mayank Mayank, Bharanidhar Duraisamy, Rohit Mohan, Abhinav Valada),经 AITOP 多维评分管线结构化提炼。